Topology optimization of line-start synchronous reluctance machine

Abstract

This paper deals with the topology optimization of line-start synchronous reluctance machines (LSSynRM). Topology optimization can lead to a reduction in energy consumption thanks to the higher efficiency of optimized electrical machines. The topology optimization is based on the normalized Gaussian network (NGnet) within this paper. The dependency of the NGnet on the distribution of Gaussian functions is analysed in detail. The evaluation algorithm of a single individual is developed. The algorithm manages communication between Ansys Maxwell and software tool SyMSpace. Furthermore, the algorithm leads to a reduction in computational time due to the preselection of unfeasible geometries. The latter is necessary due to a time-consuming transient analysis considering the machine’s start-up. The topology optimization of LSSynRM based on the NGnet is performed and the results are discussed. One selected optimized design is analysed in detail. The main contribution is the creation of the methodology for the topology optimization of LSSynRM based on the NGnet extended with the evaluation algorithm, which can be used for further improvement of line-start machines.